A combination of multi-scale calculations with machine learning for investigating hydrogen storage in metal organic frameworks

Combining multi-scale calculations with machine learning, we investigate how the ligand functionalization affects the hydrogen storage profile of Metal Organic Frameworks. The binding energy of hydrogen with 58 strategically selected functionalized benzenes was calculated with accurate ab-initio methods. Our results show that many functional groups (e.g. -OPO3H2, -OCONH2) increase the interaction strength up to 15-25% compared to benzene while -OSO3H holds the most promise with an enhancement up to 80%. Grand Canonical Monte-Carlo calculations with interatomic potentials derived from the ab-initio calculations, verify the trend obtained from the meticulous screening. In addition, a proof of principle Machine Learning analysis is performed on the ab-initio results showing a good prediction of the H-2 binding energies even with a limited amount of data. The results from our bottom-up approach lead us to conclude that this functionalization strategy can be applied to various porous materials in order to enhance their hydrogen storage performance. (C) 2021 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

Automated summary

By combining multi-scale calculations with machine learning, this study investigates how ligand functionalization affects the hydrogen storage performance of Metal Organic Frameworks. The results show that certain functional groups significantly enhance the interaction strength with hydrogen. The use of ab-initio calculations and machine learning provides a promising approach for predicting and improving hydrogen storage properties in porous materials.